Experimental Measurements of Thermoelectric Phenomena in

Nanoparticle Liquid

Suspensions (Nanofluids)

by

Moxuan Zhu

A Thesis Presented in Partial Fulfillment of the Requirements for the Degree

Master of Science

Approved November 2010 by the Graduate Supervisory Committee:

Patrick Phelan, Chair

Steven Trimble Ravi Prasher

ARIZONA STATE UNIVERSITY

December 2010

i

ABSTRACT

This study analyzes the thermoelectric phenomena of nanoparticle

suspensions, which are composed of liquid and solid nanoparticles that

show a relatively stable Seebeck coefficient as bulk solids near room

temperature. The approach is to explore the thermoelectric character of the

nanoparticle suspensions, predict the outcome of the experiment and

compare the experimental data with anticipated results. In the experiment,

the nanoparticle suspension is contained in a 15cm*2.5cm*2.5cm glass

container, the temperature gradient ranges from 20 °C to 60 °C, and room

temperature fluctuates from 20 °C to 23°C. The measured nanoparticles

include multiwall carbon nanotubes, aluminum dioxide and bismuth

telluride. A temperature gradient from 20 °C to 60 °C is imposed along the

length of the container, and the resulting voltage (if any) is measured. Both

heating and cooling processes are measured. With three different

nanoparticle suspensions (carbon nano tubes, Al2O3 nanoparticles and

Bi2Te3 nanoparticles), the correlation between temperature gradient and

voltage is correspondingly 8%, 38% and 96%. A comparison of results

calculated from the bulk Seebeck coefficients with our measured results

indicate that the Seebeck coefficient measured for each suspension is

much more than anticipated, which indicates that the thermophoresis effect

could have enhanced the voltage. Further research with a closed-loop

system might be able to affirm the results of this study.

ii

TABLE OF CONTENTS Page

LIST OF TABLES .......................................................................................... v

LIST OF FIGURES ....................................................................................... vi

CHAPTER

1 INTRODUCTION ................................................................................ 1

2 LITERATURE REVIEW OF THERMOELECTRIC MATERIALS ....... 7

A. Thermoelectric properties of bismuth telluride ............................. 7

B. The Seebeck coefficient of bismuth telluride ............................... 8

C. Solution for stable nano fluid suspension .................................. 11

3 EXPERIMENTAL DESIGN AND SETUP ......................................... 12

A. Materials used for the experiment .............................................. 12

a. Multi-walled carbon nano tubes ............................................................. 12

b. Aluminum oxide nanoparticles .............................................................. 13

c. Bismuth telluride nanoparticles ............................................................. 14

B. Equipment used for the experiment ........................................... 16

a. Container for the nanofluid .................................................................... 17

b. Heating Equipment ................................................................................ 19

c. Data collecting equipment ..................................................................... 21

d. Other equipment used ............................................................................ 23

e. Software used for data collection and analysis ...................................... 25

4 EXPERIMENTAL TRIAL AND RESULTS ........................................ 29

a. Performance Test for the Equipment ......................................... 29

iii

CHAPTER ............................................................................................... Page

b. Verification Test for Copper Wire ............................................... 31

5 THEORETICAL BACKGROUND OF THERMOPHORESIS ............ 36

6 NANOFLUID EXPERIMENTS .......................................................... 38

a. Experiment on solutions without nanoparticles .......................... 38

b. Experiment on CNT suspension ................................................ 42

c. Experiment on solution with Al2O3 nanoparticles ....................... 44

d. Experiment on Bi2Te3 nanoparticles suspension ....................... 46

e. Data Analysis: Least Square Method ......................................... 48

Least Square Method Result for NaCl Solution ......................................... 50

Least Square Method Result for CNT ........................................................ 51

Least Square Method result for Al2O3 ....................................................... 52

Least Square Method result for Bi2Te3 ...................................................... 53

f. Seebeck Coefficient Result for Bi2Te3 ........................................ 55

g. Experimental result for Bi2Te3 suspension without NaCl ........... 57

7 Discussion of Seebeck coefficient for Bi2Te3 Suspension .............. 59

a. NaCl’s thermophoresis effect ..................................................... 59

b. Thermoelectric characteristics of Bi2Te3 composites ................. 60

8 CONCLUSION .................................................................................. 62

REFERENCES ............................................................................................ 64

APPENDIX

A MATLAB CODE FOR DATA DEMONSTRATION ........................... 66

B OTHER EXPERIMENT RESULT FIGURES .................................... 69

iv

CHAPTER ............................................................................................... Page

CNT Suspension Run 1 .................................................................... 70

CNT Suspension Run 2 .................................................................... 71

Al2O3 Suspension Run 2 ................................................................... 73

Bi2Te3 Suspension Run 1 ................................................................. 74

Bi2Te3 Suspension Run 3 ................................................................. 76

v

LIST OF TABLES

Table Page

6-1 Parameters of NaCl Solution .......................................................... 40

6-2 Parameters of CNT Suspension ..................................................... 42

6-3 Parameters of Al2O3 Suspension ................................................. 44

6-4 Parameters of Bi2Te3 Suspension .................................................. 46

6-5 Least Square Method Parameters ................................................. 49

6-6 Least Square Calculating for NaCl Solution ............................... 50

6-7 Least Square Calculating for CNT .............................................. 51

6-8 Least Square Calculating for Al2O3 ............................................. 52

6-9 Least Square Calculating for Bi2Te3 ........................................... 53

6-10 Least Square Results for Materials Tested .................................... 54

6-11 dT and dV for Bi2Te3 Suspension (No NaCl) .................................. 57

vi

LIST OF FIGURES

Figure Page

1-1 Components of Thermoelectric Module .......................................... 3

1-2 Working Principle of Thermoelectric Module .................................. 4

1-2 Working Principle of Thermoelectric Generator .............................. 5

2-1 Seebeck coefficient for Different Thickness of Bi2Te3 Films ........... 9

2-2 Electrical Resistivity of Different Film Thickness .......................... 10

2-3 SEM Micrographs for 2.6ΜM, 4.5ΜM and 9.8ΜM Films .............. 11

3-1 Theoretical Structure and TEM Images of MWCNT ..................... 13

3-2 Theoretical Structure of Al2O3 ....................................................... 13

3-3 Theoretical Structure of Bismuth telluride ..................................... 14

3-4 TEM Image of Bismuth telluride Nano Rod .................................. 15

3-5 Diagram of the Experiment System .............................................. 16

3-6 Image of First Container with Insulation ....................................... 18

3-7 Container for Second Round Experiment ..................................... 19

3-8 Container with Electric Heat at the Bottom (Reverse Side) .......... 20

3-9 Newport 3040 Temperature Controller ......................................... 21

3-10 Mettler Toledo Electronic Scale .................................................... 24

3-11 pH Meter and Electric Conductivity Meter .................................... 25

3-12 Campbell Scientific PC200W Information ..................................... 25

3-13 PC200W Software Interfaces ....................................................... 26

3-14 Data Collected and Parameters Calculated in MS Excel ............. 27

3-15 Matlab Interface Loading the Collected Data ............................... 28

vii

Figure Page

4-1 Copper Run without Temperature Measurement ......................... 32

4-2 Copper Wire dT(C) vs. time(s) ...................................................... 33

4-3 Copper Wire dV(mV) vs. time(s) ................................................... 33

4-4 Copper Wire dV(mV) vs. dT(C) ..................................................... 34

6-1 NaCl Solution dT and dV vs Time .................................................. 40

6-2 NaCl dV(mV) vs. dT(C) ................................................................... 41

6-3 CNT dT and dV vs Time ................................................................. 40

6-4 CNT dT vs. dV ................................................................................ 43

6-5 Al2O3 dT and dV vs Time ................................................................ 44

6-6 Al2O3 dT vs. dV ............................................................................... 45

6-7 Bi2Te3 dT and dV vs Time............................................................... 46

6-8 Bi2Te3 dT vs. dV .............................................................................. 47

6-9 Least Square Method for Bi2Te3 suspension .................................. 48

6-10 Seebeck Coefficient for Bi2Te3 Suspension ................................... 55

6-11 dT and dV for Bi2Te3 Suspension (No NaCl) .................................. 57

1

CHAPTER 1 INTRODUCTION

The thermoelectric phenomena was discovered nearly two centuries

ago, first by Thomas Seebeck and Jean Peltier, which is regarded as the

basis of the modern thermoelectric industry (Tellurex, 2010). This

phenomenon indicates that a junction made from two different kinds of

materials, usually conductors, would show a flow of electrical current when

a temperature gradient is applied to it. On the other hand, Peltier found that

when electrical current is applied, the two different materials of the junction

would either absorb or release energy (in the form of heat). Nowadays, we

are able to explain the effect with energy transfer: the electrons in the

conductors are the carriers of the energy. As energy flows through the

materials, the temperature gradient causes a flow of electrons in a certain

direction (depends on the character of the material), and thus an electrical

current can be observed. By applying an electrical current through the

materials, the electrons move in the negative direction of the electric field,

and as the electrons are the carriers of the energy, the side that the

electrons move towards is heated, and on the contrary, the other side is

cooled.

Although the thermoelectric phenomenon was observed in the early

19th century, the devices that could make use of this effect were not

manufactured for many years. It was not until the mid-20th century that the

first practical application for a thermoelectric device was put into use

(Tellurex, 2010). With the development of modern science and technology,

2

especially in electronics and energy, the world stepped into the 21st century.

Meanwhile, our daily life largely relies on electronic devices and

transportation systems, and the main problem of these is that they not only

require a lot of energy to function, but have to dissipate heat they generate

into the surroundings. As scientists are working to develop devices to

conserve energy, thermoelectric devices can be an excellent choice to

provide cooling for the electronic devices and at the same time make use

of the energy that is now wasted in the form of heat. With the heat

dissipated, the thermoelectric modules would be able to generate direct

current (DC) electricity out of it, and the DC output can be stored and

redistributed to other electric devices. Thus the efficiency of the whole

electronic device would be increased and the working environment for the

device would be cooled down to a more appropriate temperature.

The most common thermoelectric module is shown as Fig. 1-1. The

module is composed of three parts; the upper and lower plates are made of

ceramic while the pellets contain two kinds of bismuth telluride

semiconductors, N-type (refers to negative) and P-type (refers to positive).

h

P

a

m

h

s

a

o

t

Fig

For

holes in th

P-type sem

and thus th

module. O

holes are

semicondu

and P-type

of direct cu

the other s

gure 1-1 Co

the P-type

he material

miconducto

he P-type m

On the cont

considered

uctor is the

e pellets ar

urrent the

side.

omponents

e semicon

l, the mate

or the holes

material be

trary, the N

d as the m

e negative

re connecte

module wo

3

s of Thermo

ductor, as

erial create

s are regard

ehaves as t

N-type mat

minority ca

element in

ed in serie

ould be co

oelectric M

electrons

es an abun

ded as the

the positive

terial show

arriers, wh

n the modu

es to make

ooled on on

Module (Tel

diffuse aw

ndance of

majority c

e charge el

ws a chara

hich means

ule. The N

sure that

ne side an

lurex, 2010

way from th

holes. In t

harge carr

lement in th

acter that t

s the N-typ

N-type pelle

with the flo

nd heated o

0)

he

he

ier

he

he

pe

ets

ow

on

4

Figure 1-2 Working Principle of Thermoelectric Module (Tellurex, 2010)

Figure 1-2 shows a typical utilization of a thermoelectric module, which

is providing cooling to some kind of electronic device (e.g. CPU, GPU). As

the thermoelectric device does not need any rotating components or space

for vaporization for a working liquid, it is relatively easy to maintain the

module and it can be very reliable. As a consequence, thermoelectric

modules can be used in places that require minimum space and deliver a

stable cooling effect while in a vibrating or sensitive environment, or in

other places that are not convenient for a refrigerant-based cooling system

(e.g. a conventional vapor-compression refrigeration system).

Thermoelectric devices can not only be used as a heating/cooling

device, but can also be utilized in power generation. As shown in Fig.1-3,

with a heat resource and a heat sink, thermoelectric devices can be used

to generate direct current and thus power electric equipment.

DCSource

HeatSink

ElectronicDevice

++ --

+-

N-typepellet P-typepellet

CeramicSubstrate Conductor

I

Carrier

c

f

o

t

b

s

c

a

h

n

r

w

Figure 1-3

Althoug

character o

focused on

of a nanop

thermoelec

benefits ov

solid with

containers

achieve ef

heat sourc

not least,

replace the

when nece

3 Working P

gh a great

of solid sem

n a fluid tha

particles an

ctrics do,

ver the solid

regards t

or tubes

fficient hea

ce/sink, com

for replac

e solid mo

essary.

Principle of

deal of res

miconducto

at could dis

nd fluid mix

the fluid

d. First of a

to geomet

that cont

at transfer

mpared wi

cement the

odules, res

5

f Thermoel

search has

ors, to our

splay therm

xture show

thermoelec

all, the fluid

try; secon

ain the flu

between

th a solid t

e refill of t

sulting in a

lectric Gen

s been don

knowledge

moelectric

the same

ctric mate

d itself can

ndly, with

uid, it wou

the therm

thermoelec

the fluid m

a shorter t

nerator (Tel

e on the th

e seldom h

properties

characteris

rial may h

be more fl

proper de

uld likely b

oelectric f

ctric eleme

may be ea

time to fix

llurex, 201

hermoelect

has resear

. If the res

stics as so

have certa

exible than

esign of th

be easier

fluid and th

ent. Last, b

asier than

the modu

0)

ric

rch

ult

olid

ain

n a

he

to

he

but

to

ule

6

This thesis mainly focuses on the characteristics of a nanofluid that is

composed of nanoparticles exhibiting some Seebeck coefficient and DI

(de-ionized) water, with some surfactant added to it. The following chapters

are included in this thesis: Chapter 2 consists of a literature review of

thermoelectric materials and nanofluid heat transfer properties, including

the characteristics of the most commonly used semiconductor

thermoelectric material, bismuth telluride and multiwall carbon nano tubes;

Chapter 3 covers the introduction of the design, equipment and

methodology used in the experiment process; Chapter 4 presents the

research and analysis of the thermoelectric phenomena observed in the

experiments; Chapter 5 presents the thermophoresis effect exhibited by an

NaCl solution; Chapter 6 performs the detailed analysis and comparison

between the experimental data and the theoretical results; and finally,

possible applications of a thermoelectric nanofluid and conclusions for the

entire thesis are given in Chapter 7, as well as recommendations for future

work.

7

CHAPTER 2 LITERATURE REVIEW OF THERMOELECTRIC MATERIALS

A. Thermoelectric properties of bismuth telluride

Bismuth telluride is one of the most commonly used thermoelectric

materials at room temperature, with a Seebeck coefficient of -150 μV/°C

and an electrical resistivity of 4×10-5 Ω m at room temperature. The

Seebeck coefficient can reach up to -287μV/°C at 54 °C, and also depends

on the thickness of the bismuth telluride film (Tan, et al., 2005).

Several characteristic parameters are required for evaluating a

thermoelectric material: the Seebeck coefficient, which represents the

potential of converting thermal energy into electricity; the electrical

resistivity/conductivity, which represents the ability to conduct electricity,

and the thermal conductivity, which indicates the ability to conduct heat

through the material. In order to evaluate the effectiveness of

thermoelectric materials, the dimensionless thermoelectric figure of merit

ZT is introduced in the following equations (DiSalvo, 1999):

2

,TS

ZT

[Eq. 2-1]

Or 2

,S

Z

[Eq. 2-2]

In which,

1

[Eq. 2-3]

8

where T is the temperature (°C), S the Seebeck coefficient (μV/ °C),

ρ the electrical resistivity (Ω.m), σ the electrical conductivity (S.m-1), and

λ the thermal conductivity (W m-1 °C -1). The T in equation (1) is multiplied

on both sides of the equation in order to convert the Z into a

nondimensional variable.

A material with greater ZT would be more suitable to be a

thermoelectric material. Usually, a material with ZT of 1 could be regarded

as a good one and the most up-to-date ZT values range from 2.5 to 3

(Walter, 2007). For a thermoelectric cooler (TEC), a larger ZT would lead

to a higher COP (Terry Hendricks, 2006). As Eq. 2-1 shows, in order to get

a relatively large ZT, S needs to be as large as possible while ρ and λ

should be minimized.

B. The Seebeck coefficient of bismuth telluride

The Seebeck coefficient and electric conductivity of bismuth

telluride (Bi2Te3) thin films is related to both the thickness of the films and

the temperature (Tan, 2005). Sine here Bi2Te3 is used in the form of

nanoparticles, potential size effects on their S values are of interest.

By sputtering Bi2Te3 onto a 7.5cm by 2.5 cm glass slide, the thin

films were deposited, with the thickness of the films being proportional to

the sputtering time (Tan, 2005).

(

t

c

t

o

t

t

c

c

a

Figu

(Tan, 2005

the film is

coefficient

three films

other at so

telluride th

thickness a

coefficient.

Figure 2

Since

coefficient

For the

area, not

ure 2-1 sho

5). It was o

s not prop

is observe

with differ

ome points.

in film is a

and tempe

.

2-1 Seebec

Fig. 2-1 s

is actually

e electrica

surprising

ows that S

observed by

portional to

ed at 54°C

rent thickne

. Figure 2-1

affecting the

erature for

ck coefficie

(Ta

shows the

-287μV/°C

al resistivity

gly thicker 9

S for Bi2Te

y the autho

o S. Altho

for the 9.8

esses are n

1 indicates

e result so

the highes

ent for Diffe

an, et al., 2

absolute v

C at 54 °C f

y, due to

r films ha

e3 depends

or, howeve

ough the

8-μm-thick

not similar.

s that the th

omehow. B

st possible

erent Thick

2005)

value of S

for the 9.8-

the differe

ave a rela

s on the fi

er, that the

peak of t

k film, the c

They even

hickness of

ut there is

e value of t

kness of Bi2

S, the large

-μm-thick f

ence of cro

atively sma

lm thickne

thickness

he Seebe

curves of t

n cross ea

f the bismu

an optimu

the Seebe

2Te3 Films

est Seebe

film.

oss-section

all electric

ess

of

eck

he

ch

uth

um

eck

eck

nal

cal

r

M

t

t

r

i

a

s

(

.

resistivity a

Meanwhile

the 2.6-μm

temperatur

reached a

Fig

The pa

images of

and 9.8μm

suggested

(Tan, et al.

.

and the th

e, the elect

m film, th

re while on

peak at 40

gure 2-2 El

aper also p

films for d

m were co

that with

., 2005)

inner ones

trical resist

he electric

n the cont

0°C and dro

lectrical Re

(Ta

presented s

ifferent thic

orrespondin

longer spu

10

s have a la

tivity is itse

cal resistiv

trary, the 9

opped afte

esistivity of

an, et al., 2

some SEM

cknesses.

ngly 900nm

uttering tim

arger one,

elf a functio

vity decre

9.8-μm film

er that (Tan

f Different F

2005)

M (Scanning

The grain

m, 1000nm

me, the gra

as shown

on of temp

eased with

m’s electric

n, 2005).

Film Thickn

g Electron

size for 2

m and 150

ain size be

n in Fig. 2-

perature. F

h increasin

cal resistiv

ness

Microscop

.6μm, 4.5μ

00nm, whi

ecame larg

-2.

For

ng

vity

pe)

μm

ch

ger

C

D

a

m

w

a

r

p

a

H

n

w

c

(

m

Figure

C. Solution

Due to th

agglomera

maintain w

ways to do

and the oth

relatively lo

provide a

able to dis

Huaqing X

nitric/sulfur

washed re

carbon na

(Huaqing X

methods b

e 2-3 SEM

n for stable

he charac

ate while m

well-distribu

o it: one is

her way is

ong period

long-term

ssipate pa

Xie and H

ric acid mix

epeatedly t

anotubes (

Xie, 2003)

e used to o

Micrograph

(Ta

e nanofluid

cter of di

mixed with

uted nanop

s to use an

to add surf

d (Huaqing

stable sus

articles so

Hohyun Lee

xture, after

till no acid

(MWCNT)

). In expe

obtain a ho

11

hs for 2.6Μ

an, et al., 2

d suspensio

fferent na

h water to

particles in

n ultrasonic

factant to t

Xie, 2003

spension, w

well in th

e, with bo

being dilut

dity shown

would rem

riments, it

omogeneou

ΜM, 4.5ΜM

2005)

on

anoparticle

o form a

the fluid, th

cator to bre

the solution

3). The form

while the la

he suspen

oil and ref

ted by disti

, the susp

main stab

is usually

us and sta

M and 9.8Μ

s, the pa

nanofluid.

here are tw

eak up the

n to make i

mer approa

atter way w

nsion. In t

flux in nitr

illed water,

pension of

le for ove

y suggeste

ble suspen

ΜM Films

articles m

In order

wo tradition

e aggregat

it stable for

ach may n

would not b

he paper

ric acid an

, filtered, an

multi-walle

er 2 mont

ed that bo

nsion.

ay

to

nal

es

r a

not

be

of

nd

nd

ed

hs

oth

12

CHAPTER 3 EXPERIMENTAL DESIGN AND SETUP

A. Materials used for the experiment

In order to test the thermoelectric character of nanoparticles, the

experiment used three different kinds of nanoparticles: multiwall carbon

nanotubes (CNT), aluminum oxide (Al2O3) nanoparticles and bismuth (III)

telluride (Bi2Te3) nano powder. Of which, the CNT is a material with high

electric conductivity and thermal conductivity, and at the same time, CNT

does exhibit a Seebeck coefficient.

a. Multi-walled carbon nano tubes

The theoretical structure and transmission electron microscopy

(TEM) of multi-walled carbon nano tubes (MWCNT) are shown in Fig. 3-1

(Reilly, 2007). The multi-walled carbon nano tube samples used in this

experiment were synthesized by the MER Corporation, the product was

produced without catalysts and there are 8-30 layers, with 6-20 nm in

diameter (the whole layer) and 1-5 μm in length (the whole layer), and the

density is 0.7g/ml. Due to the character of carbon nano tubes, it is essential

to add some surfactant into the solution and ultrasonicate before the

experiment.

e

X

t

b

b

O

Figu

On to

ethylene g

Xie, 2003).

the CNT an

b. Aluminu

Alum

but with a

Oxide, 201

and as

ure 3-1 The

p of addi

lycol could

. In the exp

nd surfacta

um oxide n

minum oxid

relatively

10).

Figure 3-2

s-sprayed A

eoretical St

(

ng surfac

d also work

periment, s

ant is used

nanoparticl

de (Al2O3)

high therm

Theoretica

Al2O3 nano

13

tructure an

(Reilly, 200

ctant and

k for a long

olution with

to test the

es

is usually r

mal conduc

al Structure

o-particles u

nd TEM Ima

07)

ultrasonica

ger and st

h volume fr

e character

regarded a

ctivity of 30

e of Al2O3

under TEM

ages of MW

ation, add

table soluti

raction of 1

r of the nan

s an electr

0 W.m-1°C-

(Mills, 200

M (A.I.Y. To

WCNT

ding acid

on (Huaqin

1.0% for bo

nofluid.

rical insulat

-1 (Aluminu

8)

ok, 2009)

or

ng

oth

tor,

um

n

f

a

n

e

t

0

n

c

t

e

c

p

2

Figure

nanopartic

found in th

as fast as

not require

extraneous

the suspen

0.75% and

nanofluid.

c. Bismuth

Bism

typical rep

efficient th

coefficient

presents t

2009).

3-2 show

cles under

e current e

CNT durin

ed. Neverth

s factors, s

nsion. In th

d 1.0% wer

h telluride

muth telluri

presentative

hermoelect

between -

he theoret

Figure 3-3

ws a theor

TEM (Tr

experiment

ng the expe

heless, in o

surfactant

he experim

re used to t

nanopartic

ide (Bi2Te3

e of a the

tric materia

224 μV/K a

tical struct

3 Theoretic

(Mate

14

retical stru

ransmission

ts that the n

erimental p

order to un

with the sa

ent, solutio

test the the

cles

3), a semico

rmoelectric

als for ref

at 298K (O

ture of bis

cal Structur

rialscientis

ucture of A

n Electron

nanopartic

process, an

nify the sus

ame volum

on with vol

ermoelectr

onductor is

c material.

frigeration,

O. YAMASH

smuth tellu

re of Bismu

st, 2009)

Al2O3 and

n Microsco

les do not

nd thus su

spension a

me fraction

lume fractio

ic characte

s usually re

. It is one

and has

HITA, 2004

uride (Mate

uth Tellurid

as-spraye

ope). It w

agglomera

urfactant w

and elimina

n is added

ons of 0.5%

eristics of th

egarded as

of the mo

a Seebe

4). Figure 3

erialscienti

de

ed

as

ate

as

ate

to

%,

he

s a

ost

eck

3-3

st,

F

r

t

m

e

v

o

c

F

Figure 3-4

reference

thermoelec

maintain a

example, t

value of -2

other kinds

conductivit

Figure 3-4

4 presents

line of 10

ctric materi

a stable v

the Seebec

287μV/°C

s of materia

ty.

TEM Imag

(A. P

a TEM im

0 nm (A.

ials in gene

value but

ck coefficie

at 54°C (

als, regard

15

ge of Bismu

Purkayastha

mage of bis

Purkayas

eral is that

can vary

ent of bism

Tan, 2005

dless of the

uth Tellurid

a, 2006)

smuth tellu

stha, 2006

the Seebe

according

muth tellurid

5). Similar

e electric co

de Nano Ro

uride nano

6). One a

eck coeffici

g to temp

de reaches

phenomen

onductivity

od

o rod, with

aspect abo

ient does n

erature. F

s a peak at

na occur f

y and therm

a

out

not

For

t a

for

mal

B

T

2

B. Equipm

The whole

1. Contain

The co

transfer

the sam

contain

2. Power s

The po

the hea

ment used f

Figure 3

system co

ner

ntainer ma

r the heat

me time, m

ner.

supply

wer supply

ater to gen

for the exp

3-5 Diagra

onsists of fi

ade from t

from the b

making it p

y is used to

nerate a st

16

eriment

m of the Ex

ive compon

hin glass s

bottom of t

ossible to

o supply a

table therm

xperimenta

nents, acco

slides mak

the contain

observe th

djustable e

mal gradien

al System

ording to F

kes it relati

ner to the f

hrough the

electric cur

nt along th

Fig. 3-5:

ively easy

fluid while

e transpare

rrent throug

he containe

to

at

ent

gh

er.

17

With a stable current, a smooth curve during the heating process is

achievable.

3. Heater

Since only heat is provided to the container, there are two kinds of

heaters that can be used for this experiment. One is an electric film

heater; the other is a P/N thermoelectric heater/cooler.

4. DAQ

DAQ refer to ‘Data Acquisition’ system. The DAQ can be any kind of

system that can transfer electric signals to the I/O signal and

communicate with computer. The DAQ used in this experiment included

an NI (National Instrument) module and Campbell Scientific modules.

5. Laptop Computer

The computer used is a Dell INSPIRON 6400 laptop. A USB to VGA

cable connects the laptop to the Campbell Scientific data logger which

is finally used for the data acquisition.

a. Container for the nanofluid

1. Container used for the first round of experiments

The container for the nanofluid is hand made from microscope glass

slides. The dimensions of the container are 150mm×25mm×25mm and

at the end of the container, two glass slides are used as handles. Figure

3-6 presents a brief look of the container and the insulation attached.

w

l

c

u

c

a

n

As the

which is 75

long, two s

container n

used for th

can be co

adhesive c

needed to

Figure 3-6

container

5mm×25m

slides need

needs to b

e adheren

orrupted by

can no long

be built aft

6 Image of

is made of

mm for eac

d to be bo

be cleaned

ce of glass

y the alcoh

ger seal th

ter about 6

First Cont

18

f microsco

ch slide, in

onded toge

thoroughl

s slides is m

hol. Due to

he containe

6 months of

ainer with

pe glass s

order to fo

ether. After

y by 70%

made of org

o the long

er, which m

f experime

Insulation

lides, the d

orm a conta

r each exp

alcohol. T

ganic mate

g term exp

means a ne

ents.

(Penny as

dimension

ainer 150m

periment, t

The adhesi

erial and th

periment, th

ew contain

Reference

of

mm

he

ve

us

he

ner

e)

2

2

t

w

c

t

b

t

t

h

O

i

p

2. Con

As F

25mm×2m

them form

Fig

In o

which wou

cover is ut

the suspen

b. Heating

Two

temperatur

the utilizati

hot side o

Otherwise,

is observed

possible to

ntainer use

Fig. 3-7 sh

mm ones

a containe

gure 3-7 Co

order to el

ld lead to t

tilized to m

nsion at a s

g Equipme

o thermoe

re on both

ion of the

of the ther

, the coolin

d that even

o obtain a

d for the se

hows, the s

and two 2

er while the

ontainer fo

iminate the

the volume

inimize the

stable situa

nt

electric he

sides of th

thermoelec

rmoelectric

ng effect wo

n with only

a tempera

19

econd roun

second con

25mm×25

e last one is

r the Seco

e evapora

e fraction c

e evaporat

ation.

eater/coole

he containe

ctric cooler

c cooler ha

ould be gre

y passive c

ature differ

nd of exper

ntainer is m

mm×2mm

s used as a

nd Round

tion during

change of th

ion of the f

ers are u

er. Neverth

r part, the

as to be r

eatly affect

cooling on

rence of o

riments

made of fo

m glass sli

a cover.

of Experim

g the heat

he suspen

fluid and th

used to

heless, as m

heat gene

removed c

ted. After a

the other s

over 40

our 150mm

ces. Five

ments

ting proces

sion, a gla

hus mainta

control t

mentioned

erated by th

continuous

a few trials

side, it is s

. The P

×

of

ss,

ass

ain

he

in

he

sly.

, it

still

P/N

t

g

t

t

e

u

h

t

o

w

c

t

m

o

t

thermoelec

greater tem

thermoelec

the heat g

ensure a p

using a th

heat one e

temperatur

outside sur

would corr

Figure 3

Dur

cycle are o

the electric

magnetic f

on the s

thermoelec

ctric heate

mperature

ctric heater

enerated o

proper cooli

ermoelectr

end of the

re for natu

rface of the

upt the hea

3-8 Contain

ing the pro

observed. I

c film hea

field that th

status of

ctric heate

r heats on

gradient i

rs can be a

on the coo

ing effect.

ric heater,

e container

ral cooling

e container

ater.

ner with El

ocess of th

t is either d

ater affectin

he electric

the nano

er is attach

20

one side

s needed

attached on

oling side n

In this case

an ordina

r and the

g. The elec

r at the bott

ectric Heat

he experim

due to the

ng the sig

current ge

oparticle s

hed to the

and cools

in the futu

n both side

needs to b

e, as show

ry electric

other end

ctric film he

tom, in cas

t at the Bot

ment, some

electric cur

nal of the

nerated m

suspension

bottom of

on the oth

ure resear

s of the co

be removed

wn in Fig. 3-

film heate

is left in

eater is atta

se that the

ttom (Reve

e noises in

rrent that fl

e thermoco

ight have c

n. As a

f the conta

her side. If

rch, two P

ntainer wh

d in order

-8, instead

er is used

the ambie

ached to t

suspensio

erse Side)

n the heatin

lows throug

ouple, or th

certain effe

solution,

ainer, with

f a

P/N

ile

to

of

to

ent

he

ns

ng

gh

he

ect

a

a

m

d

c

c

i

a

W

0

h

t

m

c

c

metal foil

during the

curve for th

In o

cooling pro

is introduce

as a feed b

With digita

0.01A. By

heating cyc

the temper

measurem

Figu

c. Data co

In th

collection

covered s

heating c

he result ap

order to co

ocess is na

ed (shown

back system

l screen an

controlling

cle would b

rature data

ment is not u

ure 3-9 New

ollecting eq

he first few

was the

urface as

cycle of the

ppears to b

ontrol the

atural cooli

in Figure 3

m, with a c

nd button, t

g the outpu

be controll

a with the

used and t

wport 3040

quipment

w experime

DAQ (d

21

a shield.

e experime

be much sm

temperatu

ing), Newp

3-9). The te

current outp

the minimu

ut electric c

ed easily.

computer,

emperatur

0 Tempera

ents, the m

data acqu

With the m

ent is large

moother.

ure of the

port 3040 T

emperature

put port an

um adjustm

current to

For the co

the built-i

re is contro

ture Contro

main equipm

uisition eq

modificatio

ely elimina

heating p

Temperatu

e controller

nd a measu

ment for the

the electric

nvenience

in port for

olled manua

oller (Newp

ment used

quipment)

on, the noi

ated and th

process (th

re Control

r is designe

urement po

e controller

c heater, th

of recordin

temperatu

ally.

port, 2009)

for the da

SCXI-130

se

he

he

ler

ed

ort.

r is

he

ng

ure

)

ata

00

22

manufactured by National Instrument, along with the desk top with a built-in

DAQ port. Two different kinds of DAQ modules were used in data

collection: SCXI-1100 and SCXI-1102 modules. SCXI-1100 module was

used for the temperature signals, and the SCXI-1102 module was used for

voltage signal collection.

According to the reply from NI engineer, the NI SCXI-1100 module has

a phenomenon which is named as ‘ghost effect’. This effect indicates that if

the module is measuring two different signals (like voltage and temperature)

at the same time, one signal would very likely be identical to the other one,

which is not the real signal. Due to this effect, the first set of data was

largely affected by the ghost effect and the experimental equipment

needed to be rebuilt with another kind of DAQ.

The Campbell Scientific CR23X was used as the DAQ for the

redesigned system and there was no ‘ghost effect’ for this module, which

indicated that the data collected would be reliable. As CR23X is not

compatible with LabView, it was required that a specific control and

monitoring software (PC200W, product of Campbell Scientific) be used to

control the module and collect the data. The Campbell Scientific module

records the data to the memory of itself and exchanges with the computer

through a cable every time the user clicks the collect data bottom, which

means it is not required to have a computer to control the data collection.

The probes for temperature measurement are T-type

thermocouples, attached to the inner side of each end in the container with

23

thermal epoxy, which would not only enable a good heat transfer from the

fluid to the thermocouple but also isolate the thermocouple from the

surrounding that may affect its accuracy. Both probes for measuring the

voltage are made of negative material of the thermocouple

(copper-nickel/constantan), and are shielded with metal foil to eliminate the

possible surrounding noise. A water bath tub is used to calibrate both

thermocouples before they are attached to the container. Calibration is

made in the range of 20 to 70, which is also the temperature gradient

range of the experiment.

A multimeter is used to check the electric resistance of the fluid. The

electric resistance of the mixtures of DI water and nanoparticles after

ultrasonication is relatively high (over 106 Ω).

d. Other equipment used

In order to make certain mixture solutions of nanofluid, an electronic

scale is used to measure the weight of the nanoparticles. The introduction

of the electronic scale (shown in Figure 3-10) is to make sure that the

volume fraction of the nanofluid is sufficiently accurate.

v

n

m

b

a

t

i

s

A sy

volume of t

In o

nanopartic

meter are

both meter

and standa

the PH val

indicates

suspension

Figure

yringe with

the DI wate

order to re

cle suspens

used (sho

rs before u

ard solution

ue is aroun

that the

n can be re

e 3-10 Met

division va

er added to

ecord the

sion prepa

own in Figu

using them

ns for PH a

nd 7 and th

resistivity

egarded as

24

ttler Toledo

alue of 1ml

o the soluti

PH value

ared, a PH

ure 3-11).

to measu

and electric

he electrica

is relative

s an insulat

o Electroni

l is used to

ion.

and elect

H meter an

Calibration

re the cha

c conductiv

al conductiv

ely large

tor.

c Scale

o specifical

tric conduc

nd electric

ns are also

racter of th

vity are utili

vity is clos

and the

ly control th

ctivity of th

conductiv

o needed f

he nanoflu

ized. Usua

e to 0, whi

nanopartic

he

he

vity

for

id,

ally

ch

cle

e

C

m

i

i

e. Softwar

1. Campb

Campb

Campbell

monitoring

information

interface u

Figure 3-1

re used for

bell Scientif

bell Scienti

Scientific

and impo

n of PC20

nder differ

Figure 3-1

1 PH Mete

r data colle

fic PC200W

ific CR23X

PC200W

orting the

00W softwa

ent operati

2 Campbe

25

er and Elec

ection and a

W

X data log

4.0.0.21 i

data. Fig

are while

ing status.

ell Scientific

ctric Condu

analysis

gger is use

s used as

gure 3-12

Fig. 3-13

c PC200W

uctivity Met

ed for dat

s special

presents

presents t

W Informatio

ter

ta collectio

software f

some bas

the softwa

on

on.

for

sic

are

P

Figu

PC200

PC200W

PC200W D

ure 3-13 PC

26

0W Main In

W Monitor

Data Collec

C200W Sof

nterface

Interface

ction Interfa

ftware Inte

ace

erfaces

2

r

v

M

p

3

f

A

m

t

2. Microso

The da

reference

voltage diff

Matlab, so

parameters

3-14 shows

Figure

Neverth

from Camp

As Excel

modificatio

the data in

oft Excel

ta collecte

temperatu

ference. In

ome pre-c

s: tempera

s.

3-14 Data

heless, the

pbell Scien

cannot re

ons (using

to a .txt file

ed from CR

re, hot en

n order to p

calculation

ature diffe

Collected

e Microsoft

ntific CR-23

ead the r

Matlab) ne

e in Excel.

27

R-23X data

nd tempera

process the

can be m

rence and

and Param

ft Excel ca

3X data log

rows and

eed to be m

a logger inc

ature, cold

e data muc

made in E

d Seebeck

meters Calc

annot proce

gger, which

columns

made befor

clude time,

end temp

ch faster a

Excel to a

k coefficien

culated in M

ess the da

h is saved

in a .dat

re processi

, data logg

perature a

nd clearer

achieve tw

nt as Figu

MS Excel

ata collecte

in a .dat fi

t file, som

ing and sa

ger

nd

in

wo

ure

ed

le.

me

ve

3

F

p

i

e

‘

‘

d

c

3. Matlab

The da

Figure 3-15

performed

F

First of

in .dat files

example, o

‘E:\My Dro

‘data=load

data into M

copy the w

Afte

ata collecte

5 shows th

.

igure 3-15

f all, the or

s, change

on this com

pbox\Thes

('10_15_B

Matlab mat

whole matrix

er processin

ed from th

he interface

Matlab Int

riginal data

current di

mputer, the

sis and Mee

i2Te3_1%_

trix. Double

x and past

ng by Exce

28

he data lo

e of Matlab

terface Loa

a collected

rectory to

directory i

eting\Thes

_NaCl_1%

e click on

te into Micr

el, the data

ogger are

while the p

ading the C

from the

where the

s

is\10_15_2

%_Sur01%_

the ‘data’

rosoft Exce

a are analy

analyzed

pre-treatme

Collected D

CR-23X da

e data file

2010’ and u

_run1.dat')

matrix in ‘

el for proce

zed in Mat

with Matla

ent of data

Data

ata logger

located. F

user can u

;’ to load th

‘Workspac

essing.

tlab.

ab.

a is

is

For

se

he

e’,

29

CHAPTER 4 EXPERIMENTAL TRIAL AND RESULTS

a. Performance Test for the Equipment

To make sure that the experimental equipment are functional and

reliable, several trials should be performed before the actual nanofluid

tests. Among which, sealing of the container, thermocouple calibration,

contact of the thermocouple and test for copper wire and DI water (distilled

water) are important ones.

1. Test for Sealing

Sealing of the container is of great importance in this experiment,

since it is crucial to control the volume fraction. That means that there

should not be any leaking in the container and a glass cover is included so

as to eliminate the evaporation of the fluid.

Test for sealing is relatively easy compared to other tests. The

container is first filled with DI water and placed on a paper towel, and then

after waiting for half an hour see if there is any visible water mark on the

paper towel. To be more specific, the container can be placed upon CuSO4

powder, if the powder does not turn blue, that indicates that the container is

properly sealed.

2. Thermocouple Calibration

The thermocouples used are T type thermocouples, which are

designed to measure the temperature between -250°C to 350°C, with a

sensitivity of 43μV/°C (Omega, 2010). The experiment is designed to

perform in the range of room temperature (about 22°C) to 70°C and the

30

digital water bath can be used to calibrate the thermocouples. The

temperature range of the water bath is from 0°C (with ice water mixture) to

100°C (with boiling water).

To calibrate the thermocouple, the water bath tub should be set to

different temperatures, from 20°C to 70°C, with an interval of 5°C. The

temperature of the water tub should be regarded as the reference

temperature and the two thermocouple temperatures should be recorded

and calibrated according to the different temperature interval.

3. Contact of Thermocouples

In order to make sure that the thermocouples are able to measure

the temperature while electrically insulated from the fluid, thermal epoxy is

introduced to maintain proper thermal conductivity and act as electric

insulator; another crucial function of the epoxy is to properly attach the

thermocouples at the bottom of the container.

To test the contact of thermocouples, a computer with certain

software and a hand-held infrared temperature thermometer are needed.

First run the software and monitor the temperature data acquired from the

two thermocouples attached to the bottom and compare with the

temperature readings of the inferred thermometer. Minor differences

indicate that the thermocouples are in a good contact with the container.

31

b. Verification Test for Copper Wire

In order to verify that the wires used to measure the voltage across

nanofluid are functional, several trial tests were performed.

The wire for measuring the voltage across the fluid is from the

negative material (copper-nickel) of the T type thermocouple, and the

copper wire used for testing is from the positive material of the T Type

thermocouple. In this way, a differential thermocouple is created; this

differential thermocouple can be regarded as a T type thermocouple. We

are able to compare the measured voltage with the data from the

manufacturer’s data base that indicates different voltage difference in

response to various temperature differences.

Experimental Plan:

1. Use connected copper wire and simulated voltage curve and show

the difference.

2. Wire marked by “+” is adhered to the hot end of the container and

connect to the positive junction of the DAQ board; the other wire

sticks to the cold end and is connected to the negative junction.

3. Attach a thermocouple on each end of the container.

4. Run the test for copper with and without temperature measurement

to make sure the signals are not affecting each other.

R

R

t

I

t

Results

The

Reference

to be:

In which S

For

So

Figure 4-

Figu

temperatur

dV(m

V)

eoretical es

Tables’ fro

is the See

T-Type the

-1 Copper

ure 4-1 is

re measur

stimation is

om Omega

ebeck coeff

ermocoupl

∆V S

Run withou

the verific

rement cha

32

s made us

a (Omega,

∆V S ∆

ficient of T-

es, S is ap

∆T 0

ut Tempera

cation test

annel plug

Time (s)

sing the ‘R

2010) and

∆T

-Type therm

pproximatel

0.04229∆T

ature Meas

t of the co

gged in; it

)

Revised Th

d the equa

mocouples

ly -0.04229

surement (

opper wire

t presents

hermocoup

ation is mad

s.

9 mV/.

(4/22/2010)

e without t

the voltag

ple

de

)

he

ge

c

m

c

e

i

change as

measurem

channels a

effect’ on t

is accurate

dT (

)

s a functio

ment channe

are plugge

he Campb

e.

Figur

on of temp

el plugged

ed in and i

ell Scientif

re 4-2 Cop

33

perature d

in, the curv

t is confirm

fic CX23 m

per Wire d

Time (s

ifference.

ve is simila

med that th

module. The

dT() vs. T

s)

With only

ar to the cu

here is no

e voltage m

Time(s)

the voltag

urve that bo

such ‘gho

measureme

ge

oth

ost

ent

t

Figu

temperatur

dV (

mV

) dV

(m

V)

Fi

Figur

ure 4-2 sh

re and vol

Ex

igure 4-3 C

re 4-4 Cop

hows the

tage chan

xperimentaCurve

34

Copper Wir

per Wire d

temperatu

nels in us

Time (s

dT ()

al

re dV(mV)

dV(mV) vs.

ure change

e. The hea

s)

)

Best LineadV=-0.042

vs. time(s)

dT()

e with tim

ating proce

ar Fit:25*dT

)

me with bo

ess lasts f

oth

for

35

about 200 seconds and the container was naturally cooled after that. As

Fig. 4-3 shows, the handbook data and the experimental data are close to

each other, and so the experimental data can be regarded as accurate.

There is another way to determine the Seebeck coefficient of the

thermocouple composed by probes measuring the voltage and the copper:

find a best linear fit in Fig. 4-4, and the slope (-0.0425 μV/) of the line is

the S of this ‘differential T-type ‘thermocouple’. In this case, it is confirmed

that the measurements for the temperature and voltage are accurate. The

dV vs. dT curve also shows a good pattern that the Seebeck coefficient

does not change much during the heating and cooling process and is a

closed loop.

The next step is to test the fluid with electrolyte in it, thus increasing

the electric conductivity of the fluid, and the observed phenomena will be

discussed in Chapter 5.

36

CHAPTER 5 THEORETICAL BACKGROUND OF THERMOPHORESIS

One of the main differences between experiments of solids and

fluids is the liquidity of fluids, which enables both conductive and

convective heat, mass, and charge transfer. Although it is expected that

some thermoelectric effects would be shown in the nanofluid, because of

the Seebeck coefficient of the nanoparticles, the thermophoresis effect

should also be taken into consideration.

Some studies showed that particles in a fluid can move due to a

temperature gradient, which is normally regarded as thermophoresis,

would be related to the thermoelectric effect in the fluid (Würger, 2009). For

an external electric field of (Würger, 2009):

E δα [Eq. 5-1]

where E is the external electric field, δα the reduced Seebeck

coefficient, T the temperature gradient and e the elementary charge.

The external electric field generated by the temperature gradient is

expressed in Eq. 5-1.

As the thermophoresis of the carriers in the solution would take

place in a temperature gradient, a thermoelectric voltage would be formed

accordingly (Sheng Zhang, 2004):

∆ ln [Eq. 5-2]

37

where ∆T δ ∙ grad T ≪ and V 0 and ∆ is the

thermoelectric voltage, the activation energy and the charge of the

ions.

For NaCl solutions at a temperature 25~75°C, the activation energy during

dissolution is -43.54 kJ/mol (Sheng Zhang, 2004), and q=1. During the

experiments reported in this thesis, the mean temperature was 45°C. Thus,

the Seebeck coefficient of the NaCl solution is (Sheng Zhang, 2004):

S2

483.8 /

In this case, the maximum voltage due to thermophoresis of the only the

NaCl solution would be:

V ∙ ∆T 483.8 / ∙ 30 14.51mV

38

CHAPTER 6 NANOFLUID EXPERIMENTS

Before nanofluid experiments, a trial with no nanoparticles in the

fluid was performed to verify that the voltage difference is due to the

nanoparticles.

Since the suspension is designed to be heated on one end and

naturally cooled on the other end, a thermal gradient is formed within the

container. As NaCl was added to the suspension to reduce the noise, and

nanoparticles were added to the DI water, it is highly possible that the

thermal gradient would lead to some thermophoresis effect on the ions in

the NaCl solution. Furthermore, as the particles would usually gain positive

or negative charge, the thermophoresis effect would lead to an electric

gradient and thus affect the voltage difference that is anticipated.

a. Experiment on solutions without nanoparticles

1. Components: 1%NaCl+1%Surfactant+DI Water (Positive)

2. Suspension characteristics:

G=32.9μs pH=6.82

3. Conditioning:

Process the solution in ultrasonicator for 15 to 30 minutes.

4. Procedure:

4.1 Transport the solution from container used for ultrasonic

conditioning to the glass channel used for experiment.

4.2 Turn the power for DAQ and power supply on.

39

4.3 Open PC200W software on laptop, connect the USB to VGA

cable, click on ‘connect’ bottom and wait for the software

returning ‘connected’ instruction.

4.4 Turn on the power supply to heat the hot end of container until

the temperature difference between the hot and cold end

reaches 30°C. Turn off the power supply and wait for the fluid to

naturally cool down to ambient temperature (the cold side does

not have forced cooling but natural cooling).

4.5 When the temperature difference between the hot end and cold

end dropped to 3°C, stop the DAQ and upload the data to

computer.

4.6 Switch the probe for voltage measurement and perform the

previous procedures again, and upload the data to the computer.

4.7 After the experiment, pour the nanoparticle suspension into a

container and transport to labeled waste container, clean the

container with 70% alcohol and pour the cleaning fluid into the

waste container. Write the name of the components in the fluid

on the tag of waste container (ex. ‘Carbon Nano Tubes’,

‘Alcohol’, etc.).

4.8 Make sure that all experiment equipment are in place, clean up

the experimental station and shut down all equipment.

The nanoparticles used in the experiment include aluminum oxide

(Al2O3), carbon nano tubes (CNT) and bismuth telluride (Bi2Te3).

t

w

N

d

i

l

Suspe

Na

Surfa

p

6.Table 6

the experim

with 1% v

NaCl was

during the

In Fig.

in the sam

line indica

Tab

ension Vo

aCl

actant

H

82 6-1shows t

ment. For a

olume frac

added to t

experimen

Figure

6-1, dT an

e graph, w

tes the dV

ble 6-1 Par

olume Frac

1.00%

1.00%

Electric Conductiv

32.9μs

the compo

all solution

ction of su

the suspen

nts.

e 6-1 NaCl

nd dV durin

where the b

V curve. T

40

rameters o

ction

4

ity

nents in th

s and susp

urfactant a

nsions to e

Solution d

ng the expe

blue line in

The highes

of NaCl Sol

Weight

438.34 mg

202 mg

Ultrasonic

1he NaCl sol

pensions, t

nd 1% vo

eliminate th

dT and dV v

eriment of

ndicates the

st tempera

lution

WVo

20

cate Durat

5 min lution that w

the volume

lume fract

he noise t

vs Time

NaCl solut

e dT curve

ature differ

Water olume

0 mL

ion

was used f

e was 20 m

tion of NaC

hat occurre

tion is show

e and the re

rence (28°

for

mL,

Cl.

ed

wn

ed

C)

o

f

s

N

e

e

d

b

s

occurred a

for natural

same time

NaCl solu

experimen

experimen

Figure

dV and dT

between d

solution is

dV (

mV

)

at around 7

cooling. T

, but show

tion is ex

t (theoret

t was abou

6-2 shows

T plot doe

V and dT.

not very st

700 s, at w

The dV cur

s a peak o

xpected to

tically), an

ut -6mV.

Figure 6-

s the dV an

es not imp

In other w

trong.

41

which point

rve present

of 4 mV at 3

show a

nd the m

2 NaCl dV(

nd dT relati

ply that the

words, the c

dT (°C

the power

ted does n

350s. As m

maximum

maximum d

(mV) vs. dT

ionship for

ere is a d

correlation

C)

r supply wa

not reach a

mentioned i

dV of -1

dV observ

T(C)

the NaCl s

direct linea

of dV and

as turned o

a peak at th

in Chapter

4mV in th

ved in th

solution. Th

ar correlatio

d dT for Na

off

he

5,

he

his

he

on

aCl

b

c

t

f

U

e

o

b. Experim

Suspensio

CNT

NaCl

Surfactan

pH

-

Table

control the

the previo

fraction o

Ultrasonica

evenly dist

obtained fo

ment on CN

Table

on Volum

1

1

nt 1

ECon

6-2 shows

variables,

us NaCl s

of CNT

ation was

tributed su

or the CNT

F

NT suspen

e 6-2 Para

me Fraction

1.00%

1.00%

1.00%

Electric nductivity

-

s the param

the only d

solution us

nanopartic

used for

uspension.

T suspensio

Figure 6-3 C

42

sion

ameters of

n

14

43

2

UltrasonDurat

15 m

meters for

difference b

sed in the

cles were

mixing all

pH and el

on.

CNT dT an

CNT Susp

Weight

42.45 mg

38.14 mg

202 mg

nicate tion

Ma

min Co

the CNT s

between th

e experime

e added

the comp

lectric cond

nd dV vs Ti

ension

Wa

anufacturer

MER orporation suspension

e CNT sus

ent is that

to the

ponents an

ductivity da

me

ater Volume

20 mL

r Mesh Size

-270 (<53μm

n. In order

spension a

1% volum

suspensio

nd forms a

ata were n

e

m) to

nd

me

on.

an

not

v

3

h

c

t

h

s

t

b

i

s

The ele

value of dV

300s. The

heated end

cooled end

the NaCl s

hot end is

Figure

suspension

temperatur

between dV

it can be

suspension

ectric heat

V for the C

positive p

d of the co

d. The CNT

solution, w

being heat

6-4 pres

n. The hea

re, while

V and dT.

determined

n by the an

er was shu

CNT suspe

probe for v

ontainer an

T suspens

with the vol

ted.

Figure

sents the

ating proce

the coolin

Bulk CNT

d how mu

nalysis in th

43

ut off at 70

ension was

voltage me

nd the neg

ion shows

ltage differ

6-4 CNT d

correlation

ess show a

ng process

is not a ty

uch dT and

he followin

0s in Fig. 6

s reached

easuremen

gative prob

a dV curv

rence mea

dT vs. dV

n of dV a

a curve tha

s shows

pical therm

d dV are

g part of th

6-3, while t

at 1000s, w

nt was atta

be was atta

ve different

asured drop

and dT fo

at is not pr

almost no

moelectric m

correlated

his chapter

the minimu

with a lag

ached to th

ached to th

t from that

pping as th

or the CN

roportional

o correlatio

material, an

in the CN

r.

um

of

he

he

of

he

NT

to

on

nd

NT

c

S

t

e

t

f

o

c. Experim

Suspensio

Al2O3

NaCl

Surfactan

pH

-

Table 6

the density

easier to a

time of ultr

for a same

obtained.

ment on so

Table

on VoluFrac

1.0

1.0

t 1.0

ElecCondu

-

6-3 contain

y of Al2O3

achieve an

rasonicatio

e amount

olution with

e 6-3 Param

ume ction

0%

0%

0%

ctric uctivity

U

-

ns the para

is not as l

evenly dis

n, the susp

of time. p

44

Al2O3 nan

meters of A

We

81.

434

202

Ultrasonicat

15 mins

ameters of

arge as ot

stributed s

pension wa

pH and ele

oparticles

Al2O3 Susp

eight

.4 mg

4 mg

2 mg

te Man

Nanote

f the Al2O3

ther nanop

suspension

as process

ectric cond

pension

Wa

nufacturer

echnologieINC

3 suspensio

particles, a

n with a rel

ed in the u

ductivity da

ater Volume

20 mL

Grade

s, 50 nm

on. Althoug

nd would b

latively sho

ultrasonicat

ata were n

e

e

m

gh

be

ort

tor

not

s

c

d

t

p

m

i

A

s

Accord

suspension

curves for

dV curve p

Figure

the curve,

process, t

maintained

indicates th

Al2O3 susp

square ana

F

ding to Fig

n does not

the NaCl s

presented s

6-6 shows

it is hard t

he correla

d at an ave

hat there’s

pension. Fu

alysis, to b

igure 6-5 A

g. 6-5, the

t show an

solution an

shows almo

Figure

s the curve

to see any

ation betwe

erage value

s almost no

urther discu

e presente

45

Al2O3 dT an

e tempera

obvious d

nd CNT sus

ost no dire

6-6 Al2O3 d

e of dT vs.

y correlatio

een dT an

e of -125m

o thermoele

ussion on t

ed later.

nd dV vs T

ature differ

difference,

spension.

ect correlati

dT vs. dV

. dV for the

n for dT a

nd dV is l

mV for the c

ectric chara

he correlat

ime

rence curv

compared

On the oth

ion with the

e Al2O3 su

nd dV. For

imited, an

cooling pro

acteristic s

tion is give

ve for Al2O

with the d

her hand, th

e dT curve

uspension.

r the heatin

nd the dV

ocess, whi

shown by th

n in the lea

O3

dT

he

.

In

ng

is

ch

he

ast

d

T

e

w

o

d. Experim

Solution

Bi2Te3

NaCl

Surfactant

pH

6.96 The pa

The pH va

electric con

Fig. 6-

which show

of dT and t

ment on Bi2

Tab

Volume

1.0

1.0

t 1.0

EleCond

30.arameters f

lue of Bi2T

nductivity is

Fi

-7 presents

ws a very s

the minimu

2Te3 nanop

ble 6-4 Para

e Fraction

00%

00%

00%

ectric uctivity

7 μs for the Bi2T

Te3 suspen

s close to t

gure 6-7 B

s the dT a

strong corr

um value fo

46

particles su

ameters of

We

155

434

202

UltrasonicDurati

15 minTe3 suspen

sion is alm

the NaCl s

Bi2Te3 dT a

and dV cu

relation for

or dV were

uspension

f Bi2Te3 Sus

eight

59 mg

4 mg

2 mg

cation on

ns nsion are p

most neutra

olution (32

nd dV vs T

urves for t

r dT and dV

obtained a

spension

Wate

2

Manufa

American presented

al (close to

2.9μs).

Time

he Bi2Te3

V. The max

at the same

er Volume

20 mL

acturer

Elements in Table 6-

7), while t

suspensio

ximum valu

e time (300

-4.

he

on,

ue

0s),

a

A

B

p

s

h

w

u

and there

As the Bi2T

Bi2Te3 sus

phenomen

Figure

suspension

heating an

will be sho

utilized in o

is an inver

Te3 is a typ

spension

non.

6-8 prese

n. As show

nd cooling

own with fu

order to fin

rse proport

pical thermo

would acc

Figure 6

ents a clea

wn in the fig

processes

urther analy

d the ‘See

47

tional relat

oelectric m

cordingly

6-8 Bi2Te3

ar correlati

gure, the c

. A specific

ysis in this

beck coeff

tionship be

material, it w

exhibit a

dT vs. dV

ion of dT

curve is alm

c correlatio

s chapter. A

ficient’ for t

etween the

was anticip

strong th

and dV fo

most linear

on between

A best line

he Bi2Te3 s

e two curve

pated that th

hermoelect

or the Bi2T

r for both th

n dT and d

ear fit can b

suspension

es.

he

ric

Te3

he

dV

be

n.

e

c

t

b

c

f

s

p

t

e. Data A

F

Acc

correlation

the plots,

between th

can be det

for the two

Due

squares m

parameters

A si

two param

Analysis: L

Figure 6-9 L

cording to F

between t

it is show

hem. By em

termined. In

separate g

e to the sim

method ca

s.

mple y=a+

eters (Leas

east Squa

Least Squa

Fig. 6-9, th

the temper

n that the

mploying th

n this case

graphs sho

mple linear

n be used

+bx model c

st Square

48

re Method

ares Metho

here is a w

rature grad

re is a rel

he best line

e, there is n

owing ∆T v

relationsh

d to decid

can be use

Method, 20

od for Bi2Te

way to show

ient and vo

atively sim

ear fit, a co

no need to

vs. time and

hip between

de the de

ed to identi

010).

e3 suspens

w whether

oltage diffe

mple linear

orrelation o

analyze th

d ∆V vs. tim

n ∆T and ∆

ependency

fy the corre

sion

there is a

erence. Fro

r relationsh

of dT and d

he correlatio

me.

∆V, the lea

of the tw

elation of th

ny

om

hip

dV

on

ast

wo

he

49

Table 6-5 Least Square Method Parameters

No. ∆T (oC) ∆V (V) x x y y

i xi yi xi* yi* x*y* x*x* y*y*

1

x∑

[EQ 6-1]

y∑

[EQ 6-2]

b∑

∑ [EQ 6-3]

a y bx [EQ 6-3]

Table 6-5 and EQs 6-1 to 6-4 present the procedure for calculation

of best linear fit and correlation between dT and dV.

50

Least Square Method Result for NaCl Solution

Table 6-6 Least Square Calculation for the NaCl Solution

No. ∆T (oC) ∆V (V) x x y y

i xı yı xi* yi* x*y* x*x* y*y*

Σ 12.0408 -8.066 -30.9424E-11 -5.65276E-10 20929.02 158883.2 4395.631

Table 6-2 shows the calculation parameters for the least square

results of CNT.

Thus,

a=-9.65183

b=-0.131726

For the correlation (Correlation, 2010):

cosθ∗ ∗

‖ ‖‖ ‖20929.02

√158883.2√4395.6310.791953

This shows that the correlation/dependence between ∆T and ∆V for

NaCl is 79.2%. As mentioned in Chapter 5, the voltage difference of NaCl

solution is mainly due to the effect of thermophoresis, which also presents

the thermoelectric property in the form of Seebeck coefficient.

51

Least Squares Method Results for CNT

Table 6-7 Least Square Calculation for CNT

No. ∆T (oC) ∆V (V) x x y y

i xı yı xi* yi* x*y* x*x* y*y*

Σ 7.441 -38.492 -5.7678E-11 -4.4711E-10 -292280 690023.5 855472.6

Table 6-2 shows the calculation parameters for the least squares

results of CNT.

Thus,

a=-35.3407

b=-0.4235

For the correlation (Correlation, 2010):

cosθ∗ ∗

‖ ‖‖ ‖292280

√690023.5√855472.60.38042

This shows that the correlation/dependence between ∆T and ∆V for

the CNT suspension is -0.38042. According to the character of carbon

nano tubes (CNT), which has strong electric conductivity and thermal

conductivity, the correlation between temperature and voltage gradient is

not anticipated to be very strong (38%).

52

Least Squares Method Results for Al2O3

Table 6-8 Least Square Calculation for Al2O3

No. ∆T (oC) ∆V (V) x x y y

i xı yı xi* yi* x*y* x*x* y*y*

Σ 13.24814 -126.886 -6.92E-11 -2.13E-10 -11439.4 212314.8 85164.85

Table 6-3 shows the calculation parameters for the least squares

results of Al2O3.

Thus,

a=-131.384

b=0.3396

For the correlation:

cosθ∗ ∗

‖ ‖‖ ‖11439.4

√212314.8√85164.850.08507

This shows that the correlation/dependence between ∆T and ∆V for

Al2O3 suspension is -0.08507, which indicates that the correlation of

temperature and voltage gradient is 8.5%. For bulk Al2O3 is not a typical

thermoelectric material, it is not surprised that the correlation between ∆T

and ∆V for Al2O3 suspension is low.

53

Least Squares Method Results for Bi2Te3

Table 6-9 Least Square Calculation for Bi2Te3

No. ∆T

(oC) ∆V (V) x x y y

i xı yı xi* yi* x*y* x*x* y*y*

Σ 9.494 -144.982 -1.676E-11 -3.8E-10 -91489.9 82767.9 108205.7

Table 6-3 shows the calculation parameters for the least squares

results for Al2O3.

Thus,

a=-134.487

b=-1.10538

For the correlation:

cosθ∗ ∗

‖ ‖‖ ‖91489.9

√82767.9√108205.70.96676

This shows that the correlation/dependence between ∆T and ∆V is

0.96676. As a suspension of a typical thermoelectric material, the strong

correlation between ∆T and ∆V for bismuth telluride powder is very

reasonable.

54

Table 6-10 Least Square Results for All Materials Tested

Material Correlation Thermoelectric Explanation

NaCl Solution 79.2% No Thermophoresis

CNT Suspension 38.0% Some Conductor

Al2O3 Suspension 8.5% No Insulator

Bi2Te3 Suspension 96.7% Yes Thermoelectric

Material

As a conclusion, Table 6-10 shows the least square results for all

materials tested in these experiments. The thermophoresis effect presents

the NaCl solution with a 79.2% correlation; 38.0% correlation for dT and dV

was shown for the CNT suspension, while the bulk CNT is commonly

regarded as conductor; for the Al2O3 suspension, only 8.5% correlation

was shown, which is due to the characteristic of the bulk Al2O3, which is

usually regarded as an insulator; as a bulk material with strong

thermoelectric characteristic, Bi2Te3 suspension shows a very strong

correlation (96.7%) for dT and dV.

f

c

r

t

c

F

c

S

f. Seebec

Acc

correlation

regarded a

the Bi2Te3

characteris

Figure 6-10

Com

The

F

Nev

coefficient

Seebeck c

ck Coefficie

cording to

of dT and

as typical t

3 nanopart

stic, which

0 shows, th

mpare this

erefore, the

Figure 6-10

vertheless,

is not the

coefficient

ent Result

the resu

dV for CN

thermoelec

ticle suspe

is worth c

here is a b

dV 1

equation to

e Seebeck

0 Seebeck

as ment

Seebeck c

of the su

Best LdV=-1

55

for Bi2Te3

lts from t

NT and Al2O

ctric fluid s

ension sh

comparing

est fit for th

1.1054 ∗ dT

o a standa

dV S ∗ d

coefficient

Coefficien

tioned in

coefficient o

spension

Linear fit: .1054*dT-13

the least

O3 are low

suspension

ows very

with the b

he dV vs. d

T 134.49

rd thermoe

dT

for this line

t for Bi2Te3

the introd

of the susp

and probe

34.49

squares m

and thus s

ns. On the

strong th

ulk Bi2Te3

dT curve fo

electric equ

ear fit is -1

3 Suspensi

duction, th

pension, bu

e wire tha

method, th

should not b

other han

hermoelect

material. A

or Bi2Te3:

uation:

.1054mV/K

ion

his Seebe

ut the over

t is used

he

be

nd,

ric

As

K.

eck

rall

to

56

measure the voltage difference. The probe material is from the negative

side of a T-type thermocouple, a copper-nickel alloy named constantan,

with a Seebeck coefficient of -35μV/K (eFunda: Theory of Thermocouples,

2010).

S

Then the experimental Seebeck coefficient for the Bi2Te3

suspension would be:

, 1.1404mV/K

According to Chapter 1, the Seebeck coefficient of bulk Bi2Te3 is

-287μV/K. Assuming that the Seebeck coefficient of the suspension is

proportional to the volume fraction of nanoparticles, the anticipated

Seebeck coefficient of the Bi2Te3 suspension would be:

, ∗ Vol% 287μV/K ∗ 1% 2.87μV/K

It is obvious that the theoretical Seebeck coefficient is much smaller

than the experimental value (only 0.24% of the experimental Seebeck

coefficient), which indicates that the assumption for nanoparticle Seebeck

coefficient does not apply.

Although it is pointed out that the thermophoresis effect in the fluid

would affect the thermoelectric effect that’s shown (Chapter 5), but the

influence is not as big as it is shown in the experiment. Further research is

needed for the explanation of this phenomenon and might be an effect that

is worth exploring.

g

s

H

S

c

a

a

e

e

i

o

g. Experim

As

suspension

However,

Seebeck

coefficient,

added was

F

Acc

achieved a

experimen

experimen

in the expe

of the susp

mental resu

mentioned

ns in order

due to

coefficient

, another

s performed

Figure 6-11

cording to F

at the same

t went on

ts. The ad

eriments b

pension as

ult for Bi2Te

d in the pr

r to elimina

outstandin

of Bi2Te

experimen

d.

dT and dV

Fig. 6-10, th

e time. And

n, which w

ddition of N

ut at the s

well.

57

e3 suspens

revious ch

ate the noi

ng differen

e3 suspen

nt on a B

V for Bi2Te3

he turning

d the dV cu

was also

NaCl greatl

same time

sion withou

hapters, the

ises obser

nce betwe

nsion and

Bi2Te3 sus

3 Suspens

points of th

urve exhib

observed

y eliminate

changed t

ut NaCl

e NaCl wa

rved in the

een the e

anticipate

pension w

ion (No Na

he two curv

ited some

in other

ed the nois

he Seebec

as added

experimen

experimen

ed Seebe

without Na

aCl)

ves were n

noise as th

suspensio

se measure

ck coefficie

to

nt.

tal

eck

aCl

not

he

on

ed

ent

58

For the Bi2Te3 suspension without NaCl, the best linear fit is:

dV 0.29038 ∗ dT 3.1358

Compare this equation to the standard thermoelectric equation:

dV S ∗ dT

Therefore, the Seebeck coefficient for this linear fit is -290.38μV/K.

As shown in the previous section, the Seebeck coefficient of Bi2Te3

suspension with NaCl is -1.1404mV/K and the Seebeck coefficient of

Bi2Te3 suspension without NaCl is -290.38μV/K, which is very close to the

bulk material’s Seebeck coefficient -287μV/K (Tan, et al., 2005).

It is clear that the NaCl in the suspension somehow magnified the

Seebeck coefficient to -1.1404mV/k, with the rest of the parameters

controlled to be the same. In the next chapter, a possible explanation for

this phenomenon is discussed.

59

CHAPTER 7 Discussion of Seebeck Coefficient for Bi2Te3 Suspension

As shown in the previous chapter, the Bi2Te3 suspension with NaCl

presents a very high Seebeck coefficient of -1.1404mV/K, while the

Seebeck coefficient of the Bi2Te3 suspension without NaCl is -290.38μV/K.

In this case, the NaCl solution not only eliminated the noise in the

experiment, but also magnified the Seebeck coefficient of the suspension

as well. In this chapter, the possibility of a magnified Seebeck coefficient is

discussed in two parts: NaCl’s thermophoresis effect and the

thermoelectric characteristics of Bi2Te3 composites.

a. NaCl’s thermophoresis effect

According to the analysis and discussion in Chapter 5, the

maximum voltage difference due to the NaCl solution can be as large as

-14.51mV. Assuming that part of the voltage difference is due to the

thermophoresis effect of the NaCl solution in the Bi2Te3 suspension, the

Seebeck coefficient shown by Bi2Te3 can be calculated.

The maximum voltage difference shown in the Bi2Te3 suspension

with NaCl is -30mV. If the thermophoresis effect of NaCl solution is

eliminated, the Seebeck coefficient of the Bi2Te3 suspension would be:

,,

dT

30mV 14.51mV

32K

484μV/K

60

According to the result, the estimated Seebeck coefficient for the

Bi2Te3 suspension is -484μV/K, while the experimental result of the

Seebeck coefficient for the Bi2Te3 suspension without NaCl is -290.38μV/K,

which is 60% of the estimated Seebeck coefficient. The result of this

assumption would not be sufficient to explain the high Seebeck coefficient

observed in the experiment.

b. Thermoelectric characteristics of Bi2Te3 composites

It is shown in some research that Bi2Te3 composites show different

thermoelectric characteristics compared to bulk Bi2Te3.

According to research on the thermoelectric properties of bismuth

telluride-based alloys, the Seebeck coefficient of the (Bi, Sb) 2 (Te, Se) 3

system varies with the percentage of C60 nanocomposites (N. Gothard G.

W., 2009). It is also mentioned in the research that the C60 nanocomposites

were added to decrease ZT, and the decrease of ZT in the experiment is

that electrical conductivity decreases preferentially over lattice thermal

conductivity (N. Gothard G. W., 2009). As indicated in previous research,

the decrease of mobility leads to the decrease in electrical conductivity (N.

Gothard J. S., 2010)

.A resent research shows that a peak ZT of 1.4 can be achieved in a

p-type nanocrystalline BixSb2–xTe3 bulk alloy at 100°C, while more

significantly, this alloy maintains a ZT of 1.2 at room temperature and 0.8 at

250°C, which make it useful for both cooling and power generation (Bed

61

Poudel, 2008). The research also indicates that the p-type bismuth alloy

reaches a peak of about 230μV/K at 100°C (Bed Poudel, 2008).

From the research mentioned in the previous paragraph, a high

thermoelectric performance is shown in nanocrystalline BixSb2–xTe3 bulk

alloy. In this case, it is likely that a high thermoelectric performance can

also be observed in nanoparticles or even in some nanoparticle

suspensions (nanofluid), not only due to the thermophoresis effect of NaCl

added to the suspension, but also due to the characteristic of the

suspension itself. One of the reasons why the experimental Seebeck

coefficient of the Bi2Te3 suspension without NaCl is similar to that of the

bulk material is that the Bi2Te3 nanoparticles within the suspension were

relatively stable; the nanoparticles were in good contact with each other,

and the fluid increase the conductance between the particles; thus the

Seebeck coefficient of Bi2Te3 suspension is close to the bulk material. On

the other hand, the addition of NaCl solution might have changed the

property of the nanoparticle suspension (in both thermoelectric

characteristics and electric conductivity), and thus can lead to a

suspension with even better thermoelectric performance.

62

CHAPTER 8 CONCLUSION

The introduction of this thesis presents a brief review of the history

of thermoelectrics and the working principle of thermoelectric devices.

Based on the thermoelectric effect shown in solid materials, an assumption

is made that ‘nanoparticle liquid suspensions’ will share this characteristic

as well.

One of the objectives of this thesis work is to develop an

experimental procedure and setup proper experimental equipment to verify

the assumption. In the experiment, three parameters are measured

through the data logger: time, temperature and voltage. A verification test

on the data logger is performed and double checked with the theoretical

result of the copper wire, which would verify the performance of the data

logger, excluding the possibility of a ‘ghost effect’ shown by the NI data

acquisition block. Three different liquid suspensions (carbon nano tubes,

aluminum dioxide nano powder and bismuth telluride nano powder) are

tested in the experiment, and all are based on a 1% volume fraction NaCl

solution.

The other objective of this thesis is to analyze the results of the

experiment by comparing three different suspensions’ Seebeck coefficient

and correlation between dT and dV. Due to the unique characteristics of

three different nano particles (insulator, conductor and semiconductor), the

correlation varies from 8% for Al2O3 to 38% for CNT and 96% for Bi2Te3.

With discussion of the thermophoresis effect that existed in the NaCl

63

solution, a possible impact on voltage change across the temperature

gradient is given in chapter 6.

The experimental Seebeck coefficient of the suspension for Bi2Te3

is -1.140mV/K, which is much more than expected. It would be of some

interest if further research is focused on this phenomenon and might be

able to develop a nanoparticle suspension with high thermoelectric

performance.

There are a few works worth exploring in the future study of this

thermoelectric effect in nanoparticle suspensions: a 2D/3D voltage

gradient analysis in a 2D/3D temperature gradient; design an experiment

with a closed loop to test the output power of the suspension in certain

temperature gradient; further analysis on the relationship between

thermophoresis and thermoelectric effect in the nanoparticle suspension;

thermoelectric and heat transfer characteristics for nanoparticle

suspension. With further research on the thermoelectric effect in

nanoparticle suspensions, a flexible sensor using liquid or better

thermoelectric fluid that dissipates the heat generated in electronic devices

might be possible outcomes.

64

REFERENCES

Aluminum Oxide. (2010). Retrieved from MakeItFrom: http://www.makeitfrom.com/data/?material=Alumina

Correlation. (2010). Retrieved Nov 2010, from Wikipedia ZH: http://zh.wikipedia.org/zh/%E7%9B%B8%E5%85%B3

eFunda: Theory of Thermocouples. (2010). Retrieved Nov 2010, from eFunda: http://www.efunda.com/designstandards/sensors/thermocouples/thmcple_theory.cfm

Least Square Method. (2010). Retrieved from Wikipedia ZH: http://zh.wikipedia.org/zh/%E6%9C%80%E5%B0%8F%E4%BA%8C%E4%B9%98%E6%B3%95

A. Purkayastha, F. L.-T. (2006, Feb). Low-Temperature, Template-Free Synthesis of Single-Crystal Bismuth Telluride Nanorods. Advanced Materials, 18, 496-500.

A.I.Y. Tok, F. B. (2009). Flame Synthesis of Nanoparticles. Retrieved from Nanyang Technological University: http://www.mse.ntu.edu.sg/Research/Areas/Pages/FlameSynthesis.aspx

Bed Poudel, Q. H. (2008, May). High-Thermoelectric Performance of Nanostructured Bismuth Antimony Telluride Bulk Alloys. Science, 634-638.

DiSalvo, F. J. (1999, July 30). Thermoelectric Cooling and Power Generation. Science, 285, 703-706.

Huaqing Xie, H. L. (2003). Nanofluids containing multiwalled carbon nanotubes and their enhanced. JOURNAL OF APPLIED PHYSICS, 94, 4967.

Materialscientist. (2009, July 2). File:Bi2Te3structure.jpg. Retrieved from Wikipedia: http://en.wikipedia.org/wiki/File:Bi2Te3structure.jpg

Mills, B. (2008, June 15). File:Corundum-3D-balls.png. Retrieved June 1, 2010, from Wikipedia: http://en.wikipedia.org/wiki/File:Corundum-3D-balls.png

N. Gothard, G. W. (2009). Effect of Processing Route on the Microstructure and Thermoelectric Properties of Bismuth Telluride-Based Alloys. CHEMISTRY AND MATERIALS SCIENCE, 39, 1909-1913.

65

N. Gothard, J. S. (2010). Physics Status Solidi A, 207(1).

Newport. (2009). Retrieved from Newport Corperation: http://www.newport.com/

O. YAMASHITA, H. O. (2004). Effect of metal electrode on thermoelectric power in bismuth telluride compounds. JOURNAL OF MATERIALS SCIENCE, 5653.

Omega. (2010). Thermocouple Reference Tables Type T. Retrieved from OMEGA ENGINEERING, Inc.: http://www.omega.com/temperature/z/pdf/z207.pdf

Reilly, R. M. (2007). Carbon Nanotubes: Potential Benefits and Risks of Nanotechnology in Nuclear Medicine. Journal of Nuclear Medicine, 48, 1039-1042.

Sheng Zhang, J. L. (2004, June). Dissolution kinetics of galena in acid NaCl solutions at 25—75 °C. Applied Geochemistry, 835-841.

Tan, J., Kalantar-zadeh, K., Wlodarski, W., Bhargava, S., Akolekar, D., Holland, A., et al. (2005). Thermoelectric properties of bismuth telluride thin films deposited by radio frequency magnetron sputtering. Proc. of SPIE, Vol. 5836, 711-718.

Tellurex. (2010). An Introduction to Thermoelectrics. Retrieved 2010, from http://www.tellurex.com/technology/design-manual.php

Terry Hendricks, W. T. (2006). Engineering Scoping Study of Thermoelectric Generator Systems for Industrial Waste Heat Recovery.

Walter, K. (2007, May). A Quantum contribution to Technology. Retrieved from Science and Technology Review: https://www.llnl.gov/str/May07/Williamson.html

Wan, B. Y., Wang, W. L., Liao, K. J., & Lv, J. W. (2005). Seebeck Effect in Carbon Nanotube Films. International Journal of Modern Physics, 19, 643-645.

Würger, J. M. (2009). Thermophoresis at a charged surface: the role of hydrodynamic slip. JOURNAL OF PHYSICS: CONDENSED MATTER, 21.

66

APPENDIX A

MATLAB CODE FOR DATA DEMONSTRATION

67

(1)

clear all;

close all;

data=load('10_15_Bi2Te3_1%_NaCl_1%_Sur1%_run3(flipped).txt');

time=data(:,1);

time1=time(1:285);

time2=time(286:end);

dv=data(:,8);

dv1=dv(1:285);

dv2=dv(286:end);

dt=data(:,9);

dt1=dt(1:285);

dt2=dt(286:end);

s=data(:,10);

figure(1);plot(time1,dt1,'r');hold on;plot(time2,dt2,'b');

figure(2);plot(time1,dv1,'r');hold on;plot(time2,dv2,'b');

figure(3);plot(dt1,dv1,'r');hold on;plot(dt2,dv2,'b')

figure(4);plot(time,s);

68

(2)

clear all;

close all;

Bi=load('Bi_Seebeck.txt');

Al=load('Al_Seebeck.txt');

CNT=load('CNT_Seebeck.txt');

dT_Bi=Bi(:,1);

dT_Al=Al(:,1);

dT_CNT=CNT(:,1);

S_Bi=Bi(:,2);

S_Al=Al(:,2);

S_CNT=CNT(:,2);

plot(dT_Bi,S_Bi,'r');

hold on;

plot(dT_Al,S_Al,'b');

plot(dT_CNT,S_CNT,'g');

APPENDIX B

OTHER EXPERIMENT RESULT FIGURES

CNTT Suspensi

Time vs.

Time vs.

ion Run 1

. dT

. dV

CNT

dT vs. d

T Suspensi

Time vs.

dV

ion Run 2

. dT

Time vs.

dT vs. d

. dV

dV

Al2OO3 Suspens

Time vs.

Time vs.

sion Run 2

. dT

. dV

Bi2Te

dT vs. d

e3 Suspens

Time vs.

dV

sion Run 1

. dT

Time vs.

dT vs. d

. dV

dV

Bi2Tee3 Suspens

Time vs.

Time vs.

sion Run 3

. dT

. dV

dT vs. ddV